雅可比矩阵及其有关的线性化
1、雅可比矩阵是什么
考虑向量方程对向量的导数:
设: f ⃗ ( y ⃗ ) = [ f 1 ( y ⃗ ) f 2 ( y ⃗ ) ⋮ f n ( y ⃗ ) ] n × 1 \vec{f}\left(\vec{y}\right)=\left[\begin{matrix}\begin{matrix}f_1\left(\vec{y}\right)\\f_2\left(\vec{y}\right)\\\end{matrix}\\\begin{matrix}\vdots\\f_n\left(\vec{y}\right)\\\end{matrix}\\\end{matrix}\right]_{n\times 1} f(y)= f1(y)f2(y)⋮fn(y) n×1 而 y ⃗ = [ y 1 y 2 ⋮ m ] m × 1 \vec{y}=\left[\begin{matrix}\begin{matrix}y_1\\y_2\\\end{matrix}\\\begin{matrix}\vdots\\m\\\end{matrix}\\\end{matrix}\right]_{m\times1} y= y1y2⋮m m×1
当其按照分子布局展开时即为雅可比矩阵:
∂ f ⃗ ( y ⃗ ) n × 1 ∂ y ⃗ m × 1 = [ ∂ f 1 ( y ⃗ ) ∂ y 1 ∂ f 1 ( y ⃗ ) ∂ y 2 … ∂ f 1 ( y ⃗ ) ∂ y m ⋮ ⋮ ∂ f n ( y ⃗ ) ∂ y 1 ⋱ ∂ f n ( y ⃗ ) ∂ y 2 ⋱ … ⋮ ⋮ ∂ f n ( y ⃗ ) ∂ y m ] n × m \frac{{\partial\vec{f}\left(\vec{y}\right)}_{n\times1}}{{\partial\vec{y}}_{m\times1}}=\left[\begin{matrix}\begin{matrix}\frac{\partial f_1\left(\vec{y}\right)}{\partial y_1}&\frac{\partial f_1\left(\vec{y}\right)}{\partial y_2}\\\end{matrix}&\begin{matrix}\ldots&\frac{\partial f_1\left(\vec{y}\right)}{\partial y_m}\\\end{matrix}\\\begin{matrix}\begin{matrix}\vdots\\\begin{matrix}\vdots\\\frac{\partial f_n\left(\vec{y}\right)}{\partial y_1}\\\end{matrix}\\\end{matrix}&\begin{matrix}\ddots\\\begin{matrix}\\\frac{\partial f_n\left(\vec{y}\right)}{\partial y_2}\\\end{matrix}\\\end{matrix}\\\end{matrix}&\begin{matrix}\begin{matrix}\\\begin{matrix}\ddots\\\ldots\\\end{matrix}\\\end{matrix}&\begin{matrix}\vdots\\\begin{matrix}\vdots\\\frac{\partial f_n\left(\vec{y}\right)}{\partial y_m}\\\end{matrix}\\\end{matrix}\\\end{matrix}\\\end{matrix}\right]_{n\times m} ∂ym×1∂f(y)n×1= ∂y1∂f1(y)∂y2∂f1(y)⋮⋮∂y1∂fn(y)⋱∂y2∂fn(y)…∂ym∂f1(y)⋱…⋮⋮∂ym∂fn(y) n×m
考虑一个例子: f ⃗ ( y ⃗ ) = [ f 1 ( y ⃗ ) f 2 ( y ⃗ ) ] = [ y 1 2 + y 2 2 + y 3 y 0 2 + 2 y 1 ] 2 × 1 \vec{f}\left(\vec{y}\right)=\left[\begin{matrix}f_1\left(\vec{y}\right)\\f_2\left(\vec{y}\right)\\\end{matrix}\right]=\left[\begin{matrix}y_1^2+y_2^2+y_3\\y_0^2+2y_1\\\end{matrix}\right]_{2\times1} f(y)=[f1(y)f2(y)]=[y12+y22+y3y02+2y1]2×1 而 y ⃗ = [ y 1 y 2 y 3 ] 3 × 1 \vec{y}=\left[\begin{matrix}\begin{matrix}y_1\\y_2\\\end{matrix}\\y_3\\\end{matrix}\right]_{3\times1} y= y1y2y3 3×1
求偏导得: f ⃗ ( y ⃗ ) = [ ∂ f 1 ( y ⃗ ) ∂ y 1 ∂ f 1 ( y ⃗ ) ∂ y 2 ∂ f 1 ( y ⃗ ) ∂ y 3 ∂ f 2 ( y ⃗ ) ∂ y 1 ∂ f 2 ( y ⃗ ) ∂ y 2 ∂ f 2 ( y ⃗ ) ∂ y 3 ] = [ 2 y 1 2 y 2 1 2 0 2 y 3 ] 2 × 3 \vec{f}\left(\vec{y}\right)=\left[\begin{matrix}\frac{\partial f_1\left(\vec{y}\right)}{\partial y_1}&\frac{\partial f_1\left(\vec{y}\right)}{\partial y_2}&\frac{\partial f_1\left(\vec{y}\right)}{\partial y_3}\\\frac{\partial f_2\left(\vec{y}\right)}{\partial y_1}&\frac{\partial f_2\left(\vec{y}\right)}{\partial y_2}&\frac{\partial f_2\left(\vec{y}\right)}{\partial y_3}\\\end{matrix}\right]=\left[\begin{matrix}2y_1&2y_2&1\\2&0&2y_3\\\end{matrix}\right]_{2\times3} f(y)=[∂y1∂f1(y)∂y1∂f2(y)∂y2∂f1(y)∂y2∂f2(y)∂y3∂f1(y)∂y3∂f2(y)]=[2y122y2012y3]2×3
2、利用雅可比进行线性化
考虑二维情况在平衡点处线性化
x ˙ 1 = f 1 ( x 1 , x 2 ) ⇒ {\dot{x}}_1=f_1\left(x_1,x_2\right)\ \ \ \ \ \Rightarrow x˙1=f1(x1,x2) ⇒ [ x 1 d ˙ x 2 d ˙ ] = [ ∂ f 1 ∂ x 1 ∂ f 1 ∂ x 2 ∂ f 2 ∂ x 1 ∂ f 2 ∂ x 2 ] x = x 0 [ x 1 d x 2 d ] \left[\begin{matrix}\dot{x_{1d}}\\\dot{x_{2d}}\\\end{matrix}\right]=\left[\begin{matrix}\frac{\partial f_1}{\partial x_1}&\frac{\partial f_1}{\partial x_2}\\\frac{\partial f_2}{\partial x_1}&\frac{\partial f_2}{\partial x_2}\\\end{matrix}\right]_{x=x_0}\left[\begin{matrix}x_{1d}\\x_{2d}\\\end{matrix}\right] [x1d˙x2d˙]=[∂x1∂f1∂x1∂f2∂x2∂f1∂x2∂f2]x=x0[x1dx2d]
x ˙ 2 = f 2 ( x 1 , x 2 ) ⇒ {\dot{x}}_2=f_2\left(x_1,x_2\right)\ \ \ \ \ \Rightarrow x˙2=f2(x1,x2) ⇒
中间的偏导矩阵为雅可比矩阵
来个例子: x ¨ + x ˙ + 1 x = 1 \ddot{x}+\dot{x}+\frac{1}{x}=1 x¨+x˙+x1=1
令 x 1 = x x_1=x x1=x x 2 = x ˙ x_2={\dot{x}} x2=x˙
得到状态空间方程: x ˙ 1 = x 2 {\dot{x}}_1=x_2 x˙1=x2
x ˙ 2 = x ¨ = 1 − 1 x − x ˙ = 1 − 1 x 1 − x 2 \dot{x}_2=\ddot{x}=1-\frac{1}{x}-\dot{x}=1-\frac{1}{x_1}-x_2 x˙2=x¨=1−x1−x˙=1−x11−x2
寻找其平衡点:令 x ˙ 1 = 0 {\dot{x}}_1=0 x˙1=0, x ˙ 2 = 0 {\dot{x}}_2=0 x˙2=0 可得:
x 10 = 1 , x 20 = 0 x_{10}=1,x_{20}=0 x10=1,x20=0
代入公式得: [ x 1 d ˙ x 2 d ˙ ] = [ 0 1 − ( − 1 x 1 2 ) − 1 ] x 0 [ x 1 d x 2 d ] = [ 0 1 1 − 1 ] [ x 1 d x 2 d ] \left[\begin{matrix}\dot{x_{1d}}\\\dot{x_{2d}}\\\end{matrix}\right]=\left[\begin{matrix}0&1\\-\left(-\frac{1}{x_1^2}\right)&-1\\\end{matrix}\right]_{x_0}\left[\begin{matrix}x_{1d}\\x_{2d}\\\end{matrix}\right]=\left[\begin{matrix}0&1\\1&-1\\\end{matrix}\right]\left[\begin{matrix}x_{1d}\\x_{2d}\\\end{matrix}\right] [x1d˙x2d˙]=[0−(−x121)1−1]x0[x1dx2d]=[011−1][x1dx2d]
此时, x 2 d ˙ = x 1 d − x 2 d \dot{x_{2d}}=x_{1d}-x_{2d} x2d˙=x1d−x2d
又因为 x d ¨ = x 2 d ˙ \ddot{x_d}=\dot{x_{2d}} xd¨=x2d˙, x d = x 1 d x_d=x_{1d} xd=x1d, x d ˙ = x 2 d \dot{x_d}=x_{2d} xd˙=x2d
所以,可得 x d ¨ + x d ˙ − x d = 0 \ddot{x_d}+\dot{x_d}-x_d=0 xd¨+xd˙−xd=0
完成线性化!